Video generation low-power-consumption optimization method and system based on edge calculation and storage medium

Abstract

The invention discloses a video generation low-power-consumption optimization method and system based on edge calculation and a storage medium, and relates to the field of video automatic generation. The method comprises the steps that related data of different data sources are acquired according to the type of a video to be generated, and data unification is conducted on the related data of different types through a data alignment method; respectively extracting and labeling semantic feature vectors of the data according to text data, video data and audio data, performing data volume calculation on the labeled semantic feature vectors, and determining a transmission speed; a transmission path is planned, and different semantic feature vectors are distributed to different edge computing nodes for optimizing a data scheduling mode in a video generation process; and respectively calculating matching degrees of different types of semantic feature vectors located on different edge calculation nodes, and completing matching of different data types according to the matching degrees, thereby optimizing video matching. According to the invention, the accuracy of data calculation is improved, and the deviation of content generation context is reduced.

Description

Technical Field

[0001] The present invention relates to the field of automatic video generation, and more specifically to a low-power optimization method, system, and storage medium for video generation based on edge computing. Background Art

[0002] With the advent of the big data and streaming media era, data videos, as a form of narrative visualization, are widely used in fields such as news, new media, education, economy, criminal investigation, and film and television. Common methods for producing data videos include using film and television tools, specific tools, programming code production, etc., but there is always a lack of efficient and powerful production tools. Currently, producing a high-quality data video still requires a large amount of time, financial resources, and material resources. Film and television works have higher visual expressiveness than conventional videos. To improve the production of data videos, the film and television animation production process is introduced as a methodology into the creation of data videos. For creators, it reduces the creation threshold of data videos to a certain extent and improves the artistic level of data videos; for viewers, it can enhance their visual perception and cognitive efficiency of data videos, which is conducive to more in-depth exploration of data information. The current video technology takes the data visualization design theory as the guide, takes the three-dimensional film and television animation production process as the entry point, takes the data video generation method as the research object, and takes the development and example production of data video tools as the test method, and systematically studies the three-dimensional data video generation method for the film and television animation production process.

[0003] However, the video generation methods in the prior art have a large amount of calculation, and there are doubts about the inaccuracy of the generated content in terms of language expression. Despite significant progress in dealing with natural language, it is still unable to fully and accurately understand and express complex language phenomena. Therefore, how to improve the accuracy of understanding and expression while reducing power consumption still requires further research by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a low-power optimization method, system, and storage medium for video generation based on edge computing, which solves the technical problems of large calculation amount and high algorithm complexity in the prior art for automatic video generation, so as to achieve the technical effects of reasonable scheduling, reducing data congestion, and reducing data latency, thereby further improving the accuracy of data calculation and reducing the deviation in the content generation context.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A low-power optimization method for video generation based on edge computing includes the following steps:

[0007] Obtain relevant data from different data sources according to the type of video to be generated, including video data, audio data, and text data, and use data alignment methods to unify the relevant data of different types;

[0008] Extract and label semantic feature vectors for the relevant data according to text data, video data, and audio data respectively, calculate the data volume of the completed labeled semantic feature vectors, and determine the transmission speed according to the different data volumes;

[0009] Plan the transmission path according to the transmission speed, and distribute different semantic feature vectors to different edge computing nodes to optimize the data scheduling method in the video generation process;

[0010] By retrieving the semantic feature vectors on different nodes, calculate the matching degrees of different types of semantic feature vectors located on different edge computing nodes respectively, and complete the matching of different data types according to the matching degree, optimizing video matching.

[0011] Optionally, for video data and audio data: convert the video data and audio data into numerical data by using the feature extraction method, align the numerical data, and remap the aligned data into audio data or video data to obtain the aligned audio data or video data;

[0012] For text data: extract entities from the text data, compare the entities with the entities in the database to obtain the belonging types, and complete the alignment of the text data according to the belonging types.

[0013] Optionally, use a neural network model to calculate the data volume of the semantic feature vectors: input the semantic feature vectors from the input layer, perform feature extraction on the semantic feature vectors through the convolutional layer, and a comparator is set in the output layer. The comparator compares the feature extraction results of different semantic feature vectors to obtain the sorting of the data volume sizes.

[0014] Optionally, distribute different semantic feature vectors to different edge computing nodes. Specifically, the switch assigns each edge service node to transmit to the edge computing node according to the data volume and the length of the transmission path, and monitors the congestion rate of each upload channel again through the upload device.

[0015] Optionally, multiple data service centers are deployed in the switch. Transmit the data that fails to upload to the preset failed edge computing node, and set a call seal for the failed edge computing node to reduce the power consumption of calling the failed data.

[0016] Optionally, when the congestion rate exceeds the preset threshold, assign the transmission tasks of the channels with the congestion rate exceeding the threshold to the channel with the lowest congestion rate.

[0017] A low-power optimization system for video generation based on edge computing, comprising:

[0018] A data acquisition module: used to obtain relevant data from different data sources according to the type of video to be generated, including video data, audio data, and text data, and use data alignment methods to unify the relevant data of different types;

[0019] A semantic feature vector annotation module: used to extract and annotate semantic feature vectors from the relevant data according to text data, video data, and audio data respectively, calculate the data volume of the completed annotated semantic feature vectors, and determine the transmission speed according to different data volumes;

[0020] An optimized data scheduling module: used to plan the transmission path according to the transmission speed, distribute different semantic feature vectors to different edge computing nodes, and optimize the data scheduling method during video generation;

[0021] An optimized video matching module: used to calculate the matching degree of different types of semantic feature vectors located on different edge computing nodes by retrieving the semantic feature vectors on different nodes, and complete the matching of different data types according to the matching degree, and optimize the video matching.

[0022] Optionally, it further includes a switch, in which multiple data service centers are deployed, the data that fails to be uploaded is transmitted to a preset failed edge computing node, and a call seal is set for the failed edge computing node to reduce the power consumption of failed data calls.

[0023] A computer storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of any one of the low-power optimization methods for video generation based on edge computing.

[0024] Through the above technical solutions, compared with the prior art, the present invention provides a low-power optimization method, system, and storage medium for video generation based on edge computing, and has the following beneficial effects:

[0025] 1. Improve data processing efficiency: By using data alignment methods to uniformly process relevant data of different types, the consistency and accuracy of the data can be ensured, providing a reliable basis for subsequent video generation. At the same time, the extraction and annotation of semantic feature vectors can accelerate the data understanding and processing process, thereby improving the overall data processing efficiency.

[0026] 2. Optimize resource allocation: This method determines the transmission speed based on the amount of data and plans the transmission path accordingly, distributing different semantic feature vectors to different edge computing nodes. This dynamic resource allocation method can ensure that each node can efficiently process the data it is responsible for, avoiding resource waste and bottlenecks.

[0027] 3. Reduce energy consumption: Edge computing reduces the distance and number of data transmissions by executing data processing tasks on devices or nodes at the network edge instead of in the cloud, thereby reducing energy consumption. In addition, by optimizing the data scheduling method and matching process, unnecessary calculations and data transmissions can be further reduced, further lowering the overall energy consumption.

[0028] 4. Improve video generation quality: By calculating the matching degree of different types of semantic feature vectors on different edge computing nodes and completing the matching of data types accordingly, the coordination and consistency among various elements during the video generation process can be ensured. This refined matching process helps improve the quality of video generation, making it more in line with the needs and expectations of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0030] Figure 1 is a schematic flowchart of the present invention;

[0031] Figure 2 is a schematic structural diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0033] An embodiment of the present invention discloses a low-power optimization method for video generation based on edge computing, as Figure 1 shown, including the following steps:

[0034] Step 1: Obtain relevant data from different data sources according to the type of video to be generated, including video data, audio data, and text data, and use the data alignment method to unify the relevant data of different types respectively;

[0035] Step 2: Extract and label semantic feature vectors for the relevant data according to text data, video data, and audio data respectively, calculate the data volume of the completed labeled semantic feature vectors, and determine the transmission speed according to the different data volumes;

[0036] Step 3: Plan the transmission path according to the transmission speed, and distribute different semantic feature vectors to different edge computing nodes to optimize the data scheduling method in the video generation process;

[0037] Step 4: By retrieving the semantic feature vectors on different nodes, calculate the matching degrees of different types of semantic feature vectors located on different edge computing nodes respectively, and complete the matching of different data types according to the matching degree from high to low to optimize video matching.

[0038] Further, in Step 1, for video data and audio data: convert the video data and audio data into numerical data by using the feature extraction method, align the numerical data, and remap the aligned data into audio data or video data to obtain the aligned audio data or video data;

[0039] For text data: extract entities from the text data, compare the entities with the entities in the database to obtain the belonging types, and complete the alignment of the text data according to the belonging types.

[0040] Further, in Step 2, use a neural network model to calculate the data volume of the semantic feature vectors: input the semantic feature vectors from the input layer, perform feature extraction on the semantic feature vectors through the convolutional layer, and a comparator is set in the output layer. The comparator compares the feature extraction results of different semantic feature vectors to obtain the sorting of the data volume sizes.

[0041] Furthermore, in this embodiment, a fully connected layer (also called a dense layer) can also be designed to flatten and connect the feature map output by the convolutional layer to this layer. Add a comparator module after the fully connected layer. This module can be a custom layer or function for comparing the feature extraction results of different semantic feature vectors. The comparator can calculate the similarity or difference degree between different vectors based on a certain metric of the feature extraction results (such as Euclidean distance, cosine similarity, etc.). According to the results of the similarity or difference degree, sort the semantic feature vectors to obtain the sorting of the data volume sizes.

[0042] Further, in step three, distribute different semantic feature vectors to different edge computing nodes, specifically: The switch assigns each edge service node to an edge computing node according to the data volume and the length of the transmission path, and monitors the congestion rate of each upload channel again through the upload device.

[0043] Furthermore, multiple data service centers are deployed in the switch. Transmit the data that fails to upload to a preset failed edge computing node, and set a call seal for the failed edge computing node to reduce the power consumption of calling failed data. When the congestion rate exceeds the preset threshold, assign the transmission tasks of the channels with the congestion rate exceeding the threshold to the channel with the lowest congestion rate.

[0044] The switch first evaluates the size of the data transmitted by each edge service node. This usually involves real-time monitoring and analysis of the length, frequency, and overall traffic of data packets. According to the different data volumes, the switch can dynamically adjust the transmission strategy to ensure efficient data transmission.

[0045] After determining the data volume, the switch needs to evaluate the advantages and disadvantages of different transmission paths. This includes factors such as the length, bandwidth, latency, and reliability of the path. Through intelligent algorithms, the switch can select the best transmission path to minimize the transmission time and cost while ensuring the integrity of the data.

[0046] Corresponding to Figure 1 the method shown, the present invention also discloses a low-power optimization system for video generation based on edge computing for the Figure 1 implementation of the method, and the specific structure is as Figure 2 shown, including:

[0047] Data acquisition module: used to acquire relevant data from different data sources according to the type of video to be generated, including video data, audio data, and text data, and use the data alignment method to unify the relevant data of different types;

[0048] Semantic feature vector annotation module: used to extract and annotate semantic feature vectors from the relevant data according to text data, video data, and audio data respectively, calculate the data volume of the completed annotated semantic feature vectors, and determine the transmission speed according to the different data volumes;

[0049] Optimized data scheduling module: used to plan the transmission path according to the transmission speed, and distribute different semantic feature vectors to different edge computing nodes to optimize the data scheduling method during the video generation process;

[0050] Optimized video matching module: It is used to calculate the matching degrees of different types of semantic feature vectors located on different edge computing nodes by retrieving the semantic feature vectors on different nodes, and complete the matching of different data types according to the high and low matching degrees to optimize video matching.

[0051] Furthermore, it further includes a switch. Multiple data service centers are deployed in the switch. The data with upload failure is transmitted to a preset failed edge computing node, and a call seal is set for the failed edge computing node to reduce the power consumption of calling failed data.

[0052] This embodiment finally discloses a computer storage medium. A computer program is stored on the computer storage medium. When the computer program is executed by a processor, the steps of any one of the disclosed low-power optimization methods for video generation based on edge computing are implemented.

[0053] In this specification, the various embodiments are described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the various embodiments can be referred to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, please refer to the description in the method part.

[0054] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A low-power optimization method for video generation based on edge computing, characterized in that: The following steps are involved: According to the type of video to be generated, relevant data from different data sources are obtained, including video data, audio data and text data, and data alignment methods are used to unify the relevant data of different types; Extract and annotate semantic feature vectors of relevant data according to text data, video data and audio data, calculate the data volume of the annotated semantic feature vectors, and determine the transmission speed according to the data volume; Plan the transmission path according to the transmission speed and distribute different semantic feature vectors to different edge computing nodes to optimize the data scheduling method in the video generation process; By retrieving the semantic feature vectors on different nodes, the matching degrees of different types of semantic feature vectors on different edge computing nodes are calculated respectively, and different data types are matched according to the matching degree to optimize video matching.

2. According to the low-power consumption optimization method for video generation based on edge computing in claim 1, it is characterized in that: For video data and audio data: convert the video data and audio data into numerical data by using a feature extraction method, align the numerical data, and remap the aligned data into audio data or video data to obtain aligned audio data or video data; For text data: extract entities from the text data, compare the entities with the entities in the database, obtain the types they belong to, and complete the alignment of the text data based on the types they belong to.

3. The low-power consumption optimization method for video generation based on edge computing according to claim 1 is characterized in that: The neural network model is used to calculate the data volume of the semantic feature vector: the semantic feature vector is input from the input layer, and the feature of the semantic feature vector is extracted through the convolution layer. The output layer is equipped with a comparator, which compares the feature extraction results of different semantic feature vectors to obtain the ranking of the data volume.

4. The low-power consumption optimization method for video generation based on edge computing according to claim 1 is characterized in that: Different semantic feature vectors are distributed to different edge computing nodes. Specifically, the switch dispatches each edge service node to transmit to the edge computing node according to the data volume and the length of the transmission path, and the congestion rate of each upload channel is re-monitored through the uploading device.

5. The low-power consumption optimization method for video generation based on edge computing according to claim 4 is characterized in that: Multiple data service centers are deployed in the switch to transmit the data that fails to be uploaded to the preset failed edge computing node, and set call sealing for the failed edge computing node to reduce the power consumption of failed data calls.

6. The low-power consumption optimization method for video generation based on edge computing according to claim 4 is characterized in that: When the congestion rate exceeds a preset threshold, the transmission task of the channel whose congestion rate exceeds the threshold is allocated to the channel with the lowest congestion rate.

7. A low-power consumption optimization system for video generation based on edge computing, characterized in that: include: Data acquisition module: used to acquire relevant data from different data sources according to the type of video to be generated, including video data, audio data and text data, and unify the relevant data of different types using data alignment methods; Semantic feature vector annotation module: used to extract and annotate semantic feature vectors of relevant data according to text data, video data and audio data, calculate the data volume of the annotated semantic feature vectors, and determine the transmission speed according to the data volume; Optimize data scheduling module: It is used to plan the transmission path according to the transmission speed, distribute different semantic feature vectors to different edge computing nodes, and optimize the data scheduling method in the video generation process; Optimized video matching module: It is used to retrieve semantic feature vectors on different nodes, calculate the matching degree of different types of semantic feature vectors on different edge computing nodes, and match different data types according to the matching degree to optimize video matching.

8. The low-power consumption optimization system for video generation based on edge computing according to claim 7 is characterized in that: It also includes a switch, in which multiple data service centers are deployed, which transmit the data that fails to be uploaded to the preset failed edge computing node, and set call sealing for the failed edge computing node to reduce the power consumption of failed data calls.

9. A computer storage medium, characterized in that The computer storage medium stores a computer program, and when the computer program is executed by the processor, the steps of a low-power consumption optimization method for video generation based on edge computing as described in any one of claims 1 to 6 are implemented.

Images